I'm very happy to write that the latest version of the IBM Classification Module has been released, as of today. This release, version 8.6,
represents a big step forward for the product as we focus on applying its
advanced content classification capabilities to problems in the ECM world. What
have we done? Lets run through some of the highlights of what's
new with version 8.6:
Content Collection, Email Archiving and Records Management
One segment of customers who are very interested in the value of automatic classification are compliance customers. With this in mind we've integrated the Classification Module directly into our new content collection and archiving product, IBM Content Collector. As customers use IBM Content Collector to archive their email or collect file system content, they can use the Classification Module to analyze those emails, attachments and documents and determine automatically what should be included in the repository, and how to organize it.
Similarly, we've integrated the Classification Module to IBM FileNet Records Manager. Now the Classification Module, through a few simple configuration steps in our workbench tool, can be used to automatically analyze, identify and declare content as records.
By applying automated, advanced content classification methods to their information, our customers can rely upon consistent classification logic, at high levels of accuracy. And they know that their automated solution will always (happily) participate -- reducing risk and increasing the effective accuracy of the classification decision making (because what good is a human based classification approach if no one actually participates? This is a blog posting for another time). With volumes of content growing every year, automating these tasks is increasingly a necessity.
Combining Rules with Context-Based Classification
We run a "Proof of Technology" program for the Classification Module. It is a day long workshop, scheduled nearby requesting customers, that lets them use the product, hands-on, as they're evaluating it. Based on this experience, I can say that our ECM customers have reacted very favorably to the accuracy, consistency and adaptability of our context-based classification methods. The system derives policies and category definitions by learning from examples. This provides high levels of accuracy without having to engage in extensive rule building exercises. The customers, when they get to actually use the product, are sold. Seeing is believing.
But our customers always have just one little rule that needs to be followed. . . or five . . hard and fast classification rules that always needed to be incorporated. Or they wanted to build their own complex business logic on top of the results coming from our context-based, statistical analysis approach. In response, we've greatly enhanced our rule-based classification capabilities and the ability to combine those rules with our context-based classification methods.
To quote our product documentation, the new release
. . . enables Boolean rules-based classification. IBM Classification Module Version 8.6 analyzes documents and e-mails . . . to identify key-words and phrases, patterns (such as account numbers and phone numbers) and words within a certain proximity of each other (such as occurrences of the phrase "Attorney General" in the same sentence as the word "California" ).
In turn, not only can version 8.6 execute rules on their
own, but these rules can be executed in combination with all of the methods of
analysis provide by the product. By being able to apply multiple methods of
analysis, in combination, to content, customers can realize even higher levels
of accuracy than possible from a single method on its own. The
combination of different analysis methods and classifications are delivered via
"decision plans" in the product. These decision plans are
created through a new set of wizards and workflows in our workbench user
interface, which brings us to the last big area of change and improvement.
Flexible Automation Options
Some organizations want fully automated classification of content. Others are not yet ready to make the jump to cutting out their users. Version 8.6 caters to both types of organizations.
The Classification Center's monitoring capabilities empower organizations to audit and review the results of automated classification, providing a chance to oversee the results of the bulk processing and provide feedback to the system.
For customers who want to help their users, without removing them entirely from
the classification process, the Classification Module can provide
"assisted classification" by automating the selection of the category
or classification, while still doing so in the user interface where the content
is authored. Classification Module integrates directly to SharePoint and Office
applications, populating metadata and categories automatically while still
providing the user the opportunity to change the recommendation themselves.
Simplicity for Usability, Consumability and Administration
We focused a good deal of time and effort on improving the usability and appearance of the product. We've simplified our user interfaces, added the new intuitive work flows for creating rules and have placed more online help in the user interfaces to reduce confusion when creating new policies, rules and analysis training sets. In the spirit of this simplification, we've combined the existing Classification Review and Classifier tools into a single, fully re-designed tool, the Classification Center. With these changes, the product is more efficient to set up and maintain.
We've also reduced the number of third party dependencies and simplified the installation proceses.
Lastly, we have placed more controls in the hands of the administrators of the system. Our system has always taken real-time feedback into account for training the system on the fly. Now this real-time feedback can be queued up, tested and rolled out to production in a more controlled manner (if so desired).
There's a lot in there. Let me know if you have questions or comments.